As people are increasingly utilizing portable computing devices, such as smart phones and tablets, to perform a variety of tasks, there can be advantages to improving the capabilities provided by these and other such devices. For example, users are increasingly using their portable computers to capture images and video, instead of using conventional digital cameras. Users are able to upload those images for processing to attempt to identify or recognize one or more objects represented in the images. Similar processes can be used for audio, video, and other such data. Using conventional approaches, full images are transmitted to a server for analysis, with each image being scanned individually. Such an approach can provide for an undesirable amount of latency in many situations. Unfortunately, processes for reducing the amount of latency also tend to decrease the accuracy of the results.